EP4683762A1

EP4683762A1 - Method of assessing the position of at least one mould level sensor and system for controlling the meniscus level in a mould

Info

Publication number: EP4683762A1
Application number: EP24702089.4A
Authority: EP
Inventors: Florian CHARRUAULT; Stefan Senge; Camile Wilbert José HOL
Original assignee: Tata Steel Ijmuiden BV
Current assignee: Tata Steel Ijmuiden BV
Priority date: 2023-03-21
Filing date: 2024-01-24
Publication date: 2026-01-28
Also published as: WO2024193873A1; KR20250163895A; JP2026509559A; CN120916855A

Abstract

The invention relates to a method of assessing the position of at least one mould level sensor in the mould of a continuous slab caster of molten metal, in particular molten steel, and to a system for controlling the meniscus level of molten metal in the mould of a continuous slab caster, and furthermore to a use and a computer program relating to the method.

Description

METHOD OF ASSESSING THE POSITION OF AT LEAST ONE MOULD LEVEL SENSOR AND SYSTEM FOR CONTROLLING THE MENISCUS LEVEL IN A MOULD

FIELD OF THE INVENTION

BACKGROUND TO THE INVENTION

In continuous slab casting, in particular of steel, the molten metal is fed from a tundish through a submerged entry nozzle (SEN) into a mould usually having an approximately rectangular horizontal cross-section. The mould is cooled, so that the molten metal solidifies at the outside, thereby forming a solidified shell. The solidified shell is extracted from the bottom of the mould and guided by rollers, while being further cooled, so that a cast slab is continuously formed. During the casting, the amount of molten metal fed into the mould is carefully controlled, wherein one aims at keeping the meniscus level of molten steel in the mould constant. To achieve this, usually a mould level sensor is positioned in the mould, and the signal of the mould level sensor is used in a feedback-loop to control the vertical position of the stopper of the submerged entry nozzle. The sensors used are often radiometric-based, such as the sensors produced by the company Berthold Technologies GmbH & Co. KG, Bad Wildbad, Germany. In such a radiometric sensor, a point source is arranged on one long side of the mould, and a long vertical X-ray sensor is arranged on the opposite long side of the mould. Depending on the height of the meniscus level, the strength of the signal detected by the sensor varies in a known relationship, thereby allowing determination of the meniscus level of molten metal in the mould.

Quality issues during continuous slab casting include surface defects, which may occur when slag is entrained within the molten metal, instead of remaining on the surface of the pool of molten metal in the mould. Other quality related issues during casting in the mould are cracks and inhomogeneous solidification. An inhomogeneous solidification around the perimeter of the mould leads to inhomogeneous internal material properties of the cast product. Therefore, a stable meniscus level is desirable for achieving a good continuous cast product quality. OBJECT OF THE INVENTION

It is therefore an object of the invention to optimize the meniscus level in the mould of a continuous slab caster, and to provide an improved system for controlling the meniscus level of molten metal, which results in a more stable meniscus level and therefore an improved quality of the cast product.

DESCRIPTION OF THE INVENTION

These objects are met or exceeded by a method of assessing the position of at least one mould level sensor in the mould of a continuous slab caster of molten metal, in particular molten steel, according to claim 1 , by a system for controlling the meniscus level of molten metal in the mould of a continuous slab caster according to claim 7, a use of such system according to claim 12 and a computer program according to claim 13.

According to a first aspect, the invention provides a method of assessing the position of at least one mould level sensor in the mould of a continuous slab caster of molten metal, in particular molten steel, wherein the method is carried out in a caster arrangement comprising a mould having a width and a thickness, the width being greater than the thickness, a submerged entry nozzle through which a liquid, e.g. molten metal or water, is fed into the mould, wherein the submerged entry nozzle is positioned at least approximately at the centre of the mould in the width direction, and a valve which regulates the flow of liquid through the submerged entry nozzle. The method comprises the steps of:

(a) arranging at least one mould level sensor to measure the meniscus level of the liquid at a first position along the width of the mould, thereby generating at least one meniscus level signal;

(b) operating the caster arrangement while controlling the valve in a feedback-loop in response to the at least one meniscus level signal;

(c) measuring the valve position and/or measuring the meniscus level at one or several positions along the width of the mould, while operating the caster arrangement for a predetermined time interval;

(d) calculating a measure for the fluctuation of the valve position and/or the meniscus level over the pre-determined time interval;

(e) arranging at least one mould level sensor to measure the meniscus level of the liquid at another position, different from the first position, along the width of the mould, thereby generating at least one further meniscus level signal, and repeating steps (b) to (d) using the at least one further meniscus level signal to control the valve; and

(f) determining an optimized position for the at least one mould level sensor by selecting the position along the width of the mould, at which the measure calculated in step (d) is minimum.

The invention has recognized that the position of the at least one mould level sensor influences the mould level stability when using the signal of the mould level sensor to control the flow of liquid into the mould, and therefore also the quality of the cast slab. Using a full-scale water model of a caster arrangement (as is done in one embodiment), it is possible to study in a safe environment mould level stability and the frequency of bubble entrainment for various control positions of at least one mould level sensor, wherein it is assumed that bubble entrainment in a water model correlates with slag entrainment in an actual continuous caster. Thus, it has been found that mould level stability is very important for the quality of the cast product. Because the meniscus level may fluctuate differently at different positions of the mould, the placement of the at least one mould level sensors is crucial in achieving the best meniscus level stability. Moreover, the invention has recognized that there is not one optimal position of the at least one mould level sensor that fits all casting parameters. Rather, it is expected that the optimal position of the at least one mould level sensor may depend on the width of the mould, and on the casting speed and the submergence depth (SD) of the submerged entry nozzle (SEN). Some continuous slab casters allow to vary the width of the mould (MW) during an ongoing uninterrupted casting process, whereby different positions of the mould level sensor may be optimal for different widths of the mould.

The invention has found that the position of the mould level sensor is important in achieving meniscus level stability and therefore the method of the invention allows to assess the position of at least one mould level sensor. Thereby, the method of the invention allows to study the dynamics of the meniscus level and its consequences on slag entrainment and surface defects using different control positions of one or several mould level sensors. The method can be carried out in a water model, e.g. a full-scale model, of a continuous caster, or in an actual continuous caster using molten metal, in particular in an industrial-scale caster such as a continuous slab caster in a plant or casting house. The invention is preferably applied to continuous casting of steel.

The method of the invention is carried out in a caster arrangement comprising a mould having a width and a thickness, the width being greater than the thickness, a submerged entry nozzle (SEN) through which liquid is fed into the mould, wherein the submerged entry nozzle is positioned at least approximately at the centre of the mould in the width direction, and a valve which regulates the flow of liquid through this SEN. The caster arrangement may be an industrial caster for molten metal, e.g. a continuous slab caster, such as a thin slab caster, in which case the liquid is molten metal, in particular molten steel. The industrial caster may include a ladle and a tundish, as is usual in the art. Alternatively, the caster arrangement may be a model caster using a different liquid, e.g., oil, water or an aqueous liquid. The caster model may also include a tundish and a liquid feeding system. The liquid may be withdrawn from the mould through a liquid outlet, to simulate the withdrawal of the solidified shell of the cast slab. It may be pumped back into the tundish above the mould. The model may be a full-scale model. It may also have a smaller scale, for example between 1 :5 and 1 :1.5. Working in a model caster allows to use a mould made of a transparent material, e.g., glass or plexiglass, which simplifies detection of the meniscus level, for example by optical detection, e.g., using a camera. Alternatively, ultrasound sensors (USS) may be used to measure the meniscus level. In an industrial caster for molten metal, the mould level may be measured by a radiometric sensor, such as offered by Berthold Technologies GmbH & Co. KG, Bad Wildbad, Germany (herein referred to as “Berthold sensor”) and described in DE_Whitepaper_Radiometrische-Messungen_DC00591 PR1-01.pdf on www.berthold.com, or by an electromagnetic sensor, such as an eddy-current sensor.

In both, model caster and industrial caster arrangement, the mould may have an approximately rectangular horizontal cross-section, wherein the longer side of the rectangle extends in the width direction and the narrow side in the thickness direction. The approximately rectangular cross section may have rounded corners. The sides of the rectangle may be straight or curved, in particular convex. The liquid may be fed into the mould through a SEN, wherein the amount of liquid is controlled by a valve. The valve may be any kind of valve which is controllable to vary the flow rate of liquid flowing into the mould. The valve may be a closure with a controllable position, e.g. a slide or pusher. The valve may be a stopper. In this case, the horizontal position of the stopper may be controlled to vary the flow rate of the liquid. The valve may open and close the SEN to thereby control the flow rate of liquid into the mould. The flow rate may be controlled in a feedback-loop in response to the at least one meniscus level signal, which is detected by the at least one mould level sensor. The control may be performed by a control unit, and using control algorithms as known in the art. Thus, if the meniscus level sinks below a pre-determined setpoint, the flow of liquid through the SEN is increased by opening the valve (further). If the meniscus level is above the setpoint, the flow of liquid is reduced by closing the valve or reducing its through flow. The valve may have several valve positions between completely open and completely closed. It may be continuously controllable between the completely open and the completely closed position. Alternatively, it may be controllable between a single open and single closed position only.

According to the inventive method, at least one mould level sensor is arranged to measure the meniscus level of the liquid at a first position along the width of the mould, thereby generating at least one meniscus level signal. This at least one mould level sensor is the one used to control the valve and thereby regulating the flow of liquid. It is placed at a first position, which is the position which is assessed in this run of the assessment method. Other positions may also be tested. The caster arrangement is then operated for a pre-determined time interval, while controlling the valve in response to the signal from the at least one mould level sensor. The predetermined time interval may be between 2 and 30 minutes, preferably between 5 and 20 minutes. The time interval should be long enough to reproducibly capture the amount of fluctuation of the valve positions and/or the meniscus level in this caster arrangement. The at least one mould level sensor may comprise one or several mould level sensors, in particular two mould level sensors.

While the caster arrangement is operated, the meniscus level at one or several positions along the width of the mould is measured. This measurement may be carried out with at least one mould level sensor, or with another sensor, e.g., an optical camera. In addition or alternatively, the valve position is measured during the pre-determined time interval. The valve position and/or the meniscus level may be measured at a plurality of time points during the pre-determined time interval. The sampling rate of this measurement, as well as the measurement of the meniscus level signal used to control the valve, may for example be between 1 and 500 Hz, preferably between 5 and 30 Hz. Data processing methods, in particular filtering methods, may be applied to the time series of valve position or meniscus level measurements, e.g., a low-pass filter may be applied. This may serve to filter out outlying measurements which do not contribute to the overall finding. A low-pass filter may also serve to approximate the measurement signal, e.g. of an ultrasound sensor (USS), to the measurement signal which is available at the casting house. The latter may be a radiometric sensor, which has a lower sampling rate than an USS.

In a next step, a measure for the fluctuation of the valve position and/or the meniscus level over the pre-determined time interval is calculated. It has been found that fluctuations in the meniscus level correlate with the amount of surface defects in the cast product. The position of the at least one mould level sensor may thereby be assessed in terms of mould level fluctuations. It has further been found that meniscus level fluctuations correlate with fluctuations of the valve position. Therefore, it is possible to measure the valve position, in addition to or instead of, the meniscus level, to determine the amount of meniscus level fluctuations during a run of the assessment method, and consequently the expected quality of the cast slab.

The measure for the fluctuation may be determined by receiving a set of values of the valve position and/or the meniscus level, measured at a series of time points during the pre-determined time interval, and calculating a measure for the amount of variation of the set of values. The set of values may be all measurement values of the valve position and/or the meniscus level which have been measured at a plurality of time points during the pre-determined time interval. The set of values may also be only a subset of these measurement values. The measure for the amount of variation of the set of values may be any measure known from statistics to determine an amount of variation among a set of values. The measure for the amount of variation may be the standard deviation. It may also be the variance, the standard error, or the interquartile range (IQR). Another measure for the amount of variation may be the maximum wave amplitude, i.e., the distance from peak to trough of a wave captured by a camera measuring the meniscus level over the complete width of the mould. The invention has shown that good results may be obtained when a standard deviation of the values obtained for the valve position and/or the meniscus level during the predetermined time interval is calculated in step (d).

The method not only assesses the position of at least one mould level sensor, but provides a way of determining an optimal position of at least one mould level sensor. This may be done by arranging the or another at least one mould level sensor at another position, different from the first position, along the width of the mould, thereby generating at least one meniscus level signal, and repeating steps (b) to (d) of operating the caster arrangement while controlling the valve in response to the at least one meniscus level signal from the mould level sensor at a different position, measuring the valve position and/or the meniscus level and calculating a measure for the fluctuation of the measured values during the pre-determine time interval. The optimal position for at least one mould level sensor may then be obtained by selecting the position along the width of the mould, at which the measure for the fluctuation is minimum. By these steps, the optimal position of a mould level sensor may be determined to reduce the meniscus level fluctuations and thereby slag entrainment, while the casting speed as well as the mould width (MW) and submergence depth (SD) of the SEN can be adjusted to match any desired casting parameters. In the prior art, the position of the mould level sensor has been chosen arbitrarily, not determined from measurements. According to an embodiment, the at least one mould level sensor comprises a first mould level sensor and a second mould level sensor, the first and second mould level sensors being positioned at different positions along the mould width. According to this embodiment, the valve is controlled in a feedback-loop in response to a combination of the signals of the first and second (and possibly third or further) mould level sensors. The inventors have found that mould level fluctuations may be reduced to a very low level when using two or more mould level sensors that work in parallel. In the prior art, only a single sensor has been used. When using two or more sensors, the valve may be controlled in response to a combination of the two or more meniscus level signals. The combination may be an algebraic function, e.g., a polynomial or linear function, in particular a linear combination, a sum, a weighted sum or a mean, e.g. arithmetic mean, of the signal values of the first and second (and possibly third or further) mould level sensors.

The positioning of the two or more mould level sensors may be varied in order to find the optimal positions for achieving the best meniscus level stability.

In one embodiment, the two mould level sensors are placed on one side of the SEN in the width direction, to thereby control the meniscus level variations.

According to an alternative embodiment, the first and second mould level sensors are arranged on opposite sides of the SEN along the width direction. This is advantageous since the inventors have found that, when only one sensor is used in the control loop, the meniscus level may behave differently on the control side and the uncontrolled side of the SEN resulting in an asymmetrical mould level. The addition of a second sensor on the other side of the SEN has been found to correct this asymmetry. When testing various sensor positions, it has been found that very stable meniscus levels were obtained when two sensors were positioned symmetrically with respect to the SEN. In other words, in this embodiment, the first and second mould level sensors are positioned at, at least approximately, equal distances to the centre of the mould. By at least approximately, it is meant that the distances are equal with a possible variation of ± 8%, preferably ± 5% and most preferred ± 2%. The centre of the mould in this regard means the centre of the mould in the width direction, which extends along the long sides of the mould.

According to an embodiment, the meniscus level is measured in step (c) at several positions along the width of the mould. The fluctuation of the meniscus level may be calculated at each position where the meniscus level is measured. This gives valuable information on the meniscus level fluctuations at different positions in the mould. The measure for the fluctuation of the meniscus level may then be a combination, in particular a polynomial or linear combination, a sum, a weighted sum or a mean such as an arithmetic mean, of the fluctuation of the meniscus level at each of these several positions. This is advantageous because it allows to study the meniscus level not only at a single point, but over the width of the mould. In this embodiment, also waves in the meniscus level profile may be assessed. In particular, the occurrence of standing or travelling waves, and in particular the amplitude of travelling waves, may be studied, which may be also influenced by the mould level sensor position. The several positions may for example be 4 to 80, preferably 8 to 20 positions. The positions may exclude that part of the mould where the SEN is situated. For practical reasons, mould level sensors are rarely placed at the position of the SEN in a casting house. In particular a radiometric sensor does not work well when the SEN obstructs the radiation from the radioactive source. The several positions may be equally spaced over the width of the mould, possibly with the exception of the part of the mould where the SEN is situated. For example, the meniscus level may be measured at eight positions, each spaced from one another by 8% to 15% of the total mould width (MW), wherein the position nearest the narrow side may for example be spaced 4-8% of the MW from the narrow side, for example 5%.

At each of the several positions, the meniscus level may be measured by a mould level sensor, for example an ultrasound sensor, an electromagnetic sensor or a radiometric sensor.

Measuring the meniscus level at several positions may in particular be done when assessing the mould level sensor position in a model caster arrangement. For example, several ultrasound sensors, in particular 4 to 16, preferably 6 to 12, e.g., 8, ultrasound sensors may be arranged along the mould width. One half of the ultrasound sensors (USS) may be positioned on either side of the SEN. In a preferred embodiment, the ultrasound sensors may be arranged in a row on either side of the SEN, in particular with equal spaces in between adjacent sensors.

According to another embodiment, the meniscus level may be measured in step (c) using a camera. This may be done using a water model of a continuous slab caster having a transparent mould. The camera may be placed e.g. on a stand and may look at the mould level horizontally, or under a slight angle from the top or from the bottom. The camera aperture, the field of view and the focus may be adjusted so that the meniscus level is visible in the captured images. To detect the meniscus level, the recorded images may be processed. For example, contrast and light intensity may be adjusted to best capture the meniscus level. This may for example be done using a min-max filter for adjusting the light intensity, and a Laplacian filter for adjusting the contrast. In an embodiment, the meniscus level across the mould width may be extracted from the images using an edge detection method, in particular an edge detection method focused to detect horizontal features. Further image processing, in particular morphological operations, may be used on the images in order to remove unwanted features. The image processing may be programmed in an algorithm, e.g., using MATLAB, which is an abbreviation of “MATrix LABoratory” and is a commercially available multi-paradigm programming language and numeric computing environment.

By measuring the meniscus level in step (c) using a camera, it is possible to capture the meniscus level over the width of the mould with a high image resolution, allowing to measure the meniscus level simultaneously at a plurality of positions spaced over the width of the mould. For example, the meniscus level may thereby be measured at 10 to 3000, preferably 100 to 1000 positions spaced over the width of the mould. The measure for the fluctuation of the meniscus level may be calculated at fewer positions along the width of the mould, e.g., at 8 to 100, preferably 20 to 50. This may be done by averaging the meniscus level as extracted from the camera images over several adjacent pixels along the width of the mould. The averaging will help to eliminate errors or outliers, which may occur when extracting the meniscus level from the camera images using image processing techniques, e.g., as described herein. The camera preferably records a time series of images at a rate of 10 to 500 images per second, preferably 20 to 60 images per second, resulting in a video.

According to an embodiment, the measure for the fluctuation calculated in step (d), which is used to assess the position of the at least one mould level sensor which is responsible for controlling the valve, is a measure for the amount of variation of the valve position. In this embodiment, the meniscus level signal from the at least one mould level sensor is used to control the valve which regulates the flow of liquid through the SEN, but the quality of the control is assessed not from the fluctuation of the meniscus level, but from the fluctuation of the valve position. It has been found that the variation in the valve position, in particular the position of a stopper which controls the flow of liquid from the tundish through the SEN, shows the same behaviour as the variation of the meniscus level. The advantage of this embodiment is that it can be carried out in an industrial caster arrangement using molten metal, e.g., in a continuous slab caster used to cast slabs of steel or other metal for commercial purposes. Thus, a model caster using another liquid is not necessary. In addition, the position of the at least one mould level sensor can be determined with more accuracy, since no differences between the model caster and the actual industrial caster exist which may falsify the results. Since the valve position is controlled, its position is known, e.g. in the control unit. Preferably, the valve position may vary between completely open and completely closed with between 0 and 2000, preferably 10 to 500 valve positions in between. The valve position may be recorded at a series of time points during the pre-determined time interval, e.g., at a rate of 5 to 100 Hz, and from this set of values, a measure for the amount of variation is calculated. The measure for the amount of variation may be the standard deviation. It may also be any other measure from statistics, such as variance, the standard error, or the IQR.

According to an embodiment, the measure for the fluctuation, both of the valve position and/or the meniscus level at any position, is calculated for a plurality of smaller time periods within the pre-determined time interval. For example, one smaller time period may comprise a subset of e.g., 50 to 500 consecutive measurement values, and a measure for the fluctuation is calculated for this subset. The smaller time periods may be between 0.5 and 20 seconds, preferably between 1 and 10 seconds long. The smaller time periods may be overlapping each other within the predetermined time interval. In other words, the smaller time period may be moved forward through the time series of measured data points like a sliding window. Alternatively, the smaller time periods are also consecutive, but with no overlap in the measurement values.

In this embodiment, the measured values for the valve position and/or the meniscus level are divided into several subsets of values, each subset collected during one of the smaller time periods, and a measure for the amount of variation of each subset of values is calculated. Thereby, a time evolution of the amount of fluctuation may be obtained. It has been found that this study of the mould level fluctuations at higher time resolution allows to detect slag entrainment at high precision. In particular, it has been shown that the measure for the amount of variation, when calculated for smaller time periods and thereby tracked over time, shows quite dominant peaks, and, by looking at videos of the meniscus level in a water model recorded by a camera, one finds that these dominant peaks correlate with the occurrence of bubble entrainment. In an industrial scale caster of molten metal, the air bubbles visible in the water model caster, e.g. after a wave in the meniscus level breaks, correspond to slag that could be entrained and then entrapped in the solidifying shell. Therefore, it is very valuable to detect fluctuations in the meniscus level or stopper positions which are high enough that bubble entrainment may occur. Therefore, the overall measure for the fluctuation of the meniscus level may be the number of dominant peaks observed when calculating a measure for the amount of variation in the meniscus level for a series of smaller time periods during the pre-determined time interval. “Dominant” may mean that the peak value is above a certain threshold value, e.g., above a certain percentage of the average value.

A measure for the variation within a smaller time period may be the total path travelled by the meniscus level or the valve position, in particular the stopper position. Thereby, the difference between the meniscus level or valve position from one measurement to the next is calculated, and the absolute values of these differences are added up for all measurements during the smaller time period.

By means of the method according to the invention, the dynamics of the meniscus level and its consequences on slag entrainment using different control positions of one or more mould level sensors may be investigated. It has been found that optical measurements of the meniscus level using a camera that can track the meniscus level over the whole mould width, may be used to advantage. The method of the invention has shown that two mould level sensors may lead to a more stable meniscus level and a more stable stopper position compared to a single control mould level sensor.

Moreover, it has been found that the control position, i.e., the position of the at least one mould level sensor, has a strong influence on the stability of the meniscus level and the valve. In particular, it has been found that controlling at inherently unstable positions along the width of the mould increased the meniscus level instabilities, while controlling where the meniscus level was inherently more stable was beneficial for the meniscus level and therefore for the stability of the valve.

Further, when using a camera to capture the meniscus level and evaluating the videos recorded by the camera, it was found that bubble entrainments, which correlate with slag entrainments in a commercial slab caster, correlate with large meniscus level fluctuations. These peaks may be identified by dividing the pre-determined time interval into shorter time periods and calculating a measure for the variation of the meniscus level during each shorter time period. For example, the measure may be a summation of the path travelled by the meniscus level during the time period. Thereby, a time evolution of meniscus level variation may be recorded. Counting the peaks in this signal may be a measure for the meniscus level stability.

According to another aspect, the invention is directed to a system for controlling the meniscus level of molten metal, preferably molten steel, in the mould of a continuous caster comprising a mould having a width and a thickness, the width being greater than the thickness, a submerged entry nozzle through which molten metal is fed into the mould, wherein the submerged entry nozzle is disposed at least approximately at the centre of the mould in the width direction, and a valve which regulates the flow of molten metal through the submerged entry nozzle. The system comprises at least a first mould level sensor and a second mould level sensor for measuring the level of molten metal in the mould, which are positioned at different positions along the width direction of the mould. The system may comprise more than two mould level sensors, e.g. three or four. The system further comprises a control unit configured for receiving at least first and second meniscus level signals from the at least first and second mould level sensors, respectively, and for calculating a combination, in particular a mean, of the at least first and second meniscus level signals, and for controlling the valve in a feedback-loop in response to the combination of the meniscus level signals.

All features and advantages described with respect to the method of the invention may also be applied to the system and vice versa. In particular, the continuous caster may have all features of the industrial caster arrangement which may be used to carry out the method, and the measurement of the meniscus level and signal processing may be carried out as described herein with respect to the method of the invention. The system of the invention is preferably suitable to carry out the method of the invention. The at least two mould level sensors may be radiometric or eddy current sensors.

In the inventive system, the valve is controlled in a feedback-loop in response to a combination, in particular a mean, of the signals of the first and second (and possible third or further) mould level sensors. The inventors have found that mould level fluctuations may be reduced to a very low level when using at least two mould level sensors that work in parallel. When using two or more sensors, the valve may be controlled in response to a combination of the two or more meniscus level signals. The combination may be an algebraic function of the two signals, e.g., a polynomial or linear function, in particular a linear combination, a sum, or a weighted summation of the two or more signals. In a preferred embodiment, the valve is controlled in response to the mean, e.g. arithmetic mean, of the two or more signals.

The system may be part of an industrial continuous slab caster of molten metal, in particular a steel caster. Typically, the molten metal is fed from a tundish into the mould. The tundish itself may be fed from a ladle. The mould of the continuous slab caster may have a width between about 800 and 1800 mm, preferably between 1000 and 1600 mm. In an embodiment, the width of the mould may be varied, in particular during the casting operation. For example, the width of the mould may be adjusted between 800 and 1800 mm, preferably between 1000 and 1500 mm. The thickness of the mould may be between 80 and 600 mm, preferably between 150 and 400 mm. The height of the mould may be between 800 and 2000 mm. The mould may be made of a copper alloy and may be cooled from the outside, so that the molten metal inside the mould will form a solidified shell while it travels down the mould, and is extracted from the bottom with the aid of rollers. The control unit may be a digital processing unit, for example a central processing unit (CPU). It may be part of a control system for controlling the operation of the continuous slab caster. For example, the control unit may be part of a computing device such as a computer, personal computer (PC), laptop or server. The computing device may have a user interface. The user interface may comprise a screen and/or input device such as a keyboard and/or a mouse, for the user to input control commands for the control unit.

According to a preferred embodiment, the positions of the first and second (and possibly further) mould level sensors have been determined by a method according to the first aspect of the invention. Thereby, mould level fluctuations may be assessed at various positions of the first and second or further mould level sensors, and the best positions may be determined. According to an embodiment, the positions have been assessed in the same mould and the same continuous caster in which the system for controlling the meniscus level is installed. The fluctuation of the valve position may have been used to determine the best sensor position. Alternatively, one may capture the signal from both sensors during the pre-determined time interval, compute a measure for the fluctuation of each signal, calculate a combination, in particular mean, of the two measures for the fluctuation, and select the positions along the width of the mould that have the lowest mean of the measures for the fluctuation of the meniscus level. Thereby, optimal positions for the first and second sensors may be determined. In particular, it is thereby possible to determine optimal positions for the first and second mould level sensors for different widths of the mould, in embodiments where the width of the mould may be adjusted. In another embodiment, the positions have been assessed in a model caster having the same or similar dimensions and casting parameters.

According to an embodiment, the first and second mould level sensors are arranged on opposite sides of the SEN along the width direction of the mould. It has been found that, if the mould level is controlled using a signal from one side of the SEN, the meniscus level may behave differently on either side, in particular, the meniscus level tends to be more unstable on the side where it is not controlled. Therefore, using two sensors positioned on either side of the SEN has been found to be beneficial for the meniscus level stability. In particular, the first and second mould level sensors may be positioned at, at least approximately, equal distances from the centre of the mould. In other words, the two sensors are positioned symmetrically on either side of the SEN, which is approximately at the centre of the mould. By “at least approximately” it is meant that there may be a variation of ± 8%, preferably ± 5%, more preferred ± 2% in the respective distances from the centre of the mould of the two sensors.

In an embodiment of the invention, the distances of the two sensors from the centre of the mould are 9% to 20%, preferably 12% to 17% of the mould width. In other words, it was found that good results were obtained when the mould level sensors were positioned close to the SEN. Also in this embodiment, the two sensors are preferably positioned at, at least approximately, equal distances from the centre of the mould.

According to an embodiment, the first and second sensors are disposed to be moveable in the width direction. This embodiment is advantageous because it allows to adjust the positions of the first and second mould level sensors for different casting parameters, e.g., casting speeds, submergence depth of the SEN and mould width. It is expected that the mould width will have a strong influence on the optimal control position of the mould level sensors, since the inherent stability of the meniscus level is expected to be a function of the meniscus level profile, which is itself a function of the mould width. The first and second mould level sensors may be mounted on a horizontal rail, which allows them to move along the width direction, e.g., by means of a cable line or a drive wheel.

Alternatively, the first and second mould level sensors may be selected from a plurality of mould level sensors which are spaced over the width of the mould. For example, the system may comprise 4 to 20, preferably 6 to 10 mould level sensors which are arranged at different positions along the width of the mould. They may not be spaced equally over the width of the mould, but may all be positioned at different positions which have been found suitable for controlling the flow of metal into the mould for different casting parameters. For example, they may be arranged at a distance from the centre of the mould of less than 50% of the width of the mould. Moreover, the plurality of sensors may be arranged, at least approximately, symmetrically around the centre of the mould, e.g., 2 to 8 mould level sensors on either side. In this embodiment, the system may select those two mould level sensors for controlling the valve, which have the best position for achieving low meniscus level fluctuations. Which two sensors are chosen may depend on the mould width, but also on the casting speed and submergence depth.

According to an embodiment, the width of the mould is configured to be adjustable during a continuous casting process. This may lead to different positions being the optimal positions for the mould level sensors. Therefore, preferably the first and second mould level sensors are configured to be automatically movable to positions along the mould width, in particular to positions on opposite sides of the SEN along the width direction of the mould. In particular, the positions to which the first and second sensors are moved have been determined as favourable for minimizing the meniscus level fluctuations and therefore the amount of slag entrainments by using an embodiment of the inventive method. For example, the optimal positions may be at distances from the centre of the mould or from 9% to 20%, preferably from 12% to 17% of the mould width. This embodiment allows to adjust the mould width and still achieve an optimal meniscus level stability and thereby product quality by reducing the amount of slag entrainments to a minimum.

According to an embodiment, the control unit is configured to filter out a characteristic frequency of meniscus level fluctuations from the signal used for controlling the valve. This is advantageous because it has been found by analysing videos of the meniscus level over the mould width captured by the camera, that the meniscus level exhibits waves travelling along the mould width. Since these waves sometimes break, leading to strong bubble entrainment and therefore presumable slag entrainment in the real caster, it is an important goal to dampen the amplitude of such waves. It has been found that the waves occur in cycles. When plotting the meniscus level over the mould width as a function of time, it was possible to detect waves travelling towards the centre of the mould on both sides of the SEN. Once a wave has reached the centre, another one starts to travel on the opposite side, thereby defining a cycle. In order to predict the frequency of this wave cycle, one may perform a power spectral analysis of the valve position and/or the meniscus level measured during the pre-determined time interval, for example of the meniscus level averaged over the mould widths, as measured by a camera. This may be done by performing a Fourier analysis of the measured signal in the time domain. Thereby, one obtains a power spectral density (PSD), which may display one or several peaks, which correlate to characteristic frequencies. It has been found that such frequencies may be found, in the caster arrangement used, between 0.05 and 0.5 Hz, preferably between 0.08 and 0.2 Hz. Therefore, these frequencies may be filtered out from the signal used for controlling the valve. This is advantageous because the waves are not decisive for the overall level of molten metal in the mould, and therefore should not be used for controlling the valve, since this might lead to an amplification of meniscus level instabilities, in particular waves.

The invention is also directed to a continuous caster for molten metal, in particular steel, comprising a system according to the invention. The continuous caster may in particular be an industrial continuous slab caster, e.g. a thin slab caster. The invention is also directed to a use of the system according to the invention in a continuous caster, in particular slab caster, for molten metal, in particular steel. All features and advantages of the method and system of the invention also apply to the continuous slab caster and the use, and vice versa.

In an alternative embodiment, the invention is directed to a system for controlling the meniscus level of a liquid in the mould of a model caster arrangement, wherein the mould has a width and a thickness, the width being greater than the thickness, a submerged entry nozzle through which liquid is fed into the mould, wherein the submerged entry nozzle is disposed at least approximately at the centre of the mould in the width direction, and a valve which regulates the flow of liquid through the submerged entry nozzle. The system comprises at least a first mould level sensor and a second mould level sensor for measuring the meniscus level of the liquid in the mould, which are positioned at different positions along the width direction of the mould. The system further comprises a control unit configured for receiving at least first and second meniscus level signals from the first and second mould level sensors, respectively, and for calculating a combination, in particular a mean, of the at least first and second meniscus level signals, and for controlling the valve in a feedback-loop in response to the combination of the meniscus level signals. In this alternative, the caster arrangement is a caster model using a liquid like water or oil. The mould level sensors may be USS or a camera. The system may be arranged for measuring the valve position and/or measuring the meniscus level at one or several along the width of the mould, while operating the caster arrangement for a pre-determined time interval, and calculating a measure for the fluctuation of the valve position and/or the meniscus level measured in step over the pre-determined time interval. The system may include a camera for measuring the meniscus level over the width of them mould, as described herein. The system may be adapted to carry out the method of the invention as described herein with regard to the model caster. All feature described herein with respect to the model caster are applicable to the system according to the alternative embodiment of the invention and vice versa. Also in this embodiment, the mould width may be adjustable.

The invention is also directed to a computer program comprising computer-executable code which, when executed by a computer, performs the steps of receiving a plurality of sets of values representing the valve position or the meniscus level at one or several positions along the width of the mould, measured at a pre-determined time interval; calculating measures for the amount of variation of each of the sets of values; and determining the set of values having the lowest measure for the amount of variation as optimal. All features and advantages of the method and system of the invention also apply to the computer program and vice versa. In particular, the computer program may perform the method of determining an optimal position of the mould level sensors by processing the measurement values of the valve position and/or meniscus level obtained from the inventive method. The computer program may be executed on any calculating unit or computer, e.g., a server, cloud computer, mobile device, laptop or PC.

The invention is also directed to a computer program product which includes the computer program. The computer program product may be supplied in a downloadable format on a server or may be provided on a digital storage medium. The invention is also directed to a non-transient digital storage medium comprising the computer program. The storage medium may be an optical, solid state or magnetic storage medium. It may for example be a hard disc, SD-card, SSD-card, USB-stick, cloud computer or any digital storage medium on a mobile device such as a laptop, tablet or mobile phone.

DETAILED DESCRIPTION OF THE FIGURES

The invention shall now be described by means of embodiments with reference to the attached drawings. In the drawings:

Fig. 1 shows a schematic cross-section through a continuous slab caster;

Fig. 2 shows a simplified view of a model caster arrangement, which may be used in an embodiment of the method;

Fig. 3 shows a side view of the mould of Fig. 2 showing the sensor positions;

Fig. 4 shows the meniscus level measured by a camera and a USS with a low pass filter applied; Fig. 5 shows the standard deviation of the meniscus level depending on the control position, using single control and mean control;

Fig. 6 shows the standard deviation of the meniscus level measured optically using single and mean control;

Fig. 7 shows the standard deviation of the stopper position for different control positions using the single control and the mean control;

Fig. 8 shows the power spectral density of the stopper position;

Fig. 9 shows the path travelled by the meniscus level, measured optically, over the predetermined time interval;

Fig. 10 is a flow diagram of a method according to an embodiment of the invention.

Fig. 1 shows a cross-section of a continuous slab caster 1 according to an embodiment of the invention. In the continuous slab caster, molten metal 8, in particular steel, is kept in a ladle 7, from which it is transferred into a tundish 6. From the tundish 6, the liquid metal is fed through a submerged entry nozzle 3 into the mould 2. The flow of molten steel is controlled by a valve 5, which in this case is a stopper, which may be moved up or down (in this embodiment), and the movement of which is controlled by the control unit 28. The molten steel forms a liquid pool 12 in the mould 2. The meniscus level of the liquid pool 12 is illustrated at 4, and a mould level sensor 10 may measure the height of the meniscus level. The sensor 10 may for example be a radiometric or an eddy-current sensor. The mould 2 may be cooled by spray cooling 18, so that the steel forms a solidified shell 16. This solidified shell 16 is withdrawn from the mould at the bottom and guided by rollers 14 into a horizontal alignment, whilst still being spray-cooled by water sprays 18. At 24, the now fully solidified strand may be cut off to form a slab 20. In other embodiments, the strand is fed straight into a hot-rolling mill, to thereby continuously producing hot-rolled steel products, e.g., steel sheets.

EXAMPLE

Fig. 2 shows part of a model caster arrangement, also referred to as thin slab caster water model, which was used to carry out experiments using a method according to an embodiment of the invention. It includes a tundish (not shown), a feeding system 3 and a mould 2 corresponding to the mould of a thin slab caster. The mould 2 is made of a transparent material such as glass or plexiglass. It has a mould width MW and a mould thickness MT. The SEN 3 is disposed approximately at the centre of the MW and reaches into the liquid 9, which may be water, at a submergence depth SD. The meniscus level 4 of the liquid 9 may be measured by ultrasound sensors USS (not shown in Fig. 2) and/or by a camera 30. The casting speed as well as the mould width and submergence depth could be adjusted to match the desired casting parameters in the casting house. The meniscus level was controlled via a feedback-loop whose input parameter is the meniscus level signal acquired by one or two ultrasound sensors (USS), which are not shown in Fig. 2. Since the USS acquire at a higher rate than the radiometric sensors in the plant, a low- pass filter was used on the signal acquired by the USS to mimic the Berthold sensor used in the plant. The camera may acquire a video, i.e. a time series of images, of at least the field of view 32, which includes a view of the meniscus level 4 over the entire mould width. The meniscus level 4 is controlled to be at the mould level setpoint illustrated at 33.

The camera 30 may be mounted on a stand (not shown), so that its height is the same as or slightly above the meniscus level 4. This allows to clearly see the meniscus level at any time and any location. An algorithm based on image processing and developed in MATLAB was used to track and measure the meniscus level.

The results shown herein were obtained using a fixed set of casting parameters. Therein, the casting speed was 5.4 m/min, the submergence depth SD was 290 mm and the mould width was 1500 mm. The pre-determined time interval was 15 minutes, in other words each measurement was performed for a duration of 15 minutes in order to achieve data convergence.

A total of eight different equidistant positions of the USS were studied. These positions are illustrated in Fig. 3, which shows a schematic side view of the thin slab caster water model mould 2. The mould width MW was 1500 mm and the SEN 3 is shown in the centre 34 of the mould 2. The eight different positions of the USS are arranged symmetrically around the centre line 36 of the mould 2. For the experiments, two USS 10a and 10b were placed at equal distances from the symmetry axis 36 of the mould, namely at one of the four symmetrical positions 1 L1 R, 2L2R, 3L3R or 4L4R. The distances between positions 1 , 2, 3 and 4 were 150 mm, the distances of positions 4L and 4R from the symmetry axis 36 were 215 mm, and the distances of positions 1 L and 1 R to the narrow sides of the mould were 85 mm. For each position (1 , 2, 3 or 4), sensors were in some experiments used independently for controlling the flow of liquid. In other words, only the left sensor “L” or the right sensor “R” was used in the control loop, while the other sensor was merely measuring the meniscus level. In other experiments, the mean signal of the two sensors was used in the control loop. These control strategies are also referred to as “single control” and “mean control” herein. In other experiments not reported herein, the two sensors were arranged not symmetrically, e.g., the sensor positions were 1 L2R, 3L4R, 4L3R, etc.

Fig. 4 illustrates the meniscus level signal measured over an interval of two minutes using the optical camera (line 38) and using the ultrasound sensor (line 40) and with a low-pass filter applied on the signal. Fig. 4 shows the meniscus level at one of the eight positions illustrated in Fig. 3. The agreement between the USS and the camera measurement is very satisfactory. The small differences between the two measurement techniques may be explained by the fact that the camera measurement is slightly less accurate than the USS. Further, the two techniques are not measuring at the exact same locations, since the USS measure at the centre of the mould in mould signal direction, while the camera measures the meniscus level on the transparent side wall.

Using the above-described setup, the standard deviation of the meniscus level was measured and calculated using different positions for the mould level sensor(s) controlling the flow of liquid (also termed “control positions”). Fig. 5 shows the standard deviation of the meniscus level depending on the control position. In the corresponding experiments, the USS 10a, 10b were positioned at four different positions arranged symmetrically around the axis 36, namely 1 L1 R, 2L2R, 3L3R and 4L4R. The caster arrangement was then operated while using either the signal of one USS 42 or using the mean signal of both sensors 44 in the control loop. The meniscus level fluctuation was obtained by calculating the standard deviation of the meniscus level measured by each of the two USS 10a, 10b over a pre-determined time interval, and then taking the mean of these two standard deviations. First of all, one can see from Fig. 5 that the standard deviation of the meniscus level fluctuations is always considerably smaller when using the mean control 44, compared to single sensor control 42. Indeed, the meniscus level fluctuations are decreased by about 15 to 25% when using the two sensors in control for each control position. Further, one can see that the maximum standard deviation is obtained at the position 3L3R. The lowest standard deviation is obtained at the position 4L4R.

Fig. 6 shows the standard deviation of the overall meniscus level measured by the camera, also using single control 45 and mean control 46. The findings are the same as with the meniscus level measured with the USS: mean control always gave better results than single control. The lowest standard deviation was obtained at the control position 4L4R, i.e., with the two ultrasound sensors at positions of 215 mm or about 14% of the mould width from the centre line 36 of the mould.

This means that the mould level stability benefits from a control position relatively close to the SEN. However, the highest level of fluctuations can be found at position 3, which appears to result from the fact that a travelling wave breaks at this location. It appears that the mould is inherently more unstable at this position.

Fig. 7 illustrates the standard deviation of the stopper position for different control positions, again using the single control 48 or the mean control 50. The bar chart follows closely the behaviour of the meniscus level fluctuations 10 in Figs. 5 and 6. This shows that the method of the invention can also be performed by assessing the position of the mould level sensor through the amount of fluctuations of the stopper position, which is highly advantageous since it allows to perform the method also in an industrial slab caster.

By analysing the meniscus level over the mould width as a function of time, as recorded by the camera, the occurrence of waves in the mould has also been analysed. In particular, waves travel towards the SEN. Once a wave has reached the centre, another one starts to travel on the opposite side of the SEN. This defines a cycle. The duration of the cycle has been observed to be about 5 to 15 seconds. The phase velocity of the waves was about 0.12 m/s.

In order to detect the frequency of this wave cycle, a power spectra analysis of the measurement data gathered by the method was performed, in particular the complete meniscus level measured by the camera, the stopper position, or the meniscus level measured at two symmetric positions by the ultrasound sensors. As an example, Fig.8 illustrates the power spectral density, normalized by the standard deviation of the signal, for the stopper position. The power spectral density (PSD) shows a characteristic frequency T, which corresponds to the frequency of a wave cycle. This frequency was about 0.11 Hz. In some experiments, additional characteristic frequencies were found corresponding to the first and second mode oscillations of a gravity wave, whose wave length is twice the mould width. Accordingly, the control loop has been designed to purposely filter out these characteristic frequencies, since gravity waves do not alter the average meniscus level. The frequency analysis may further be used to predict when entrapments originating from entrainment due to surface waves may occur in the cast slab.

Finally, Fig. 9 illustrates an embodiment, in which the measure for the fluctuation of the valve position and/or the meniscus level is calculated for a time series of smaller time periods. Thereby, a time evolution 54 of the fluctuation of meniscus level during the pre-determined time interval (in this case 900 sec = 15 min) is obtained. Fig.9 specifically illustrates this time evolution of the level of fluctuation of the meniscus level, averaged over the complete mould width, as detected by a camera. In this case, the smaller time intervals were overlapping, and the measure for the fluctuation was the sum of the absolute values of the difference between the meniscus level from one measurement point to the next (path travelled). The dashed lines show the time points, at which bubble entrainment was observed in the video recorded by the camera. The dominant peaks 56 in the signal of Fig. 9 are indicated by black triangles. One can see that the peak positions are correlated very strongly with bubble entrainment. While not all peaks lead to bubble entrainment, the peaks are nevertheless indicative of strong meniscus level fluctuations. Therefore, the number of peaks is a suitable measure for assessing the control position when using at least one mould level sensor for controlling the flow of liquid to the mould. Using this method, a strong reduction in the number of peaks could be seen when using the mean control compared to the single control, in particular the number of peaks was reduced between 20 to 50%. It could further be found that the control position that led to the most unstable meniscus level in terms of potential bubble entrainment is position 3, while positions 1 and 4 seem to be the most favourable.

Finally, Fig. 10 is an illustration of the method according to the invention. At step 60, the caster assembly is set up, and at least one, preferably two, mould level sensors are arranged at a certain position along the mould width to measure the meniscus level at that position. In step 62, the caster arrangement is operated while controlling the valve and the feedback-loop in response to the meniscus level signal from the at least one mould level sensor. In step 64, while the caster arrangement is operating in this way, the valve position and/or the meniscus level at one or several positions along the width of the mould is measured. This may be done for the meniscus level using a camera or using the at least one mould level sensor, or using even more mould level sensors. The stopper position can also be measured using the existing mechanism for controlling the stopper position.

When this is done for a pre-determined time, e.g., for 5 to 20 minutes, a measure for the fluctuation of the valve position and/or the meniscus level is calculated in step 66. This may for example be the standard deviation.

This is repeated for several different control position, i.e., different positions of the at least one mould level sensor, for as many times as there are different positions to test. This is indicated by arrow 68. Finally, once all positions have been testes, in step 70, the calculated measures for the amount of fluctuation are compared between the different control positions, and that control positions having the lowest measure of fluctuation is selected as the optimal control position.

LIST OF REFERENCE SIGNS

1 continuous slab caster

2 mould

3 submerged entry nozzle (SEN)

4 meniscus level

5 valve

6 tundish

7 ladle

8 molten steel

9 liquid

10 mould level sensor

12 liquid pool

14 support roll

16 solidified shell

18 spray cooling

20 slab

24 cutoff point

28 control unit

30 camera

32 field of view

33 mould level setpoint

MW mould width MT mould thickness

SD submergence depth

1 L, 2L, 3L, 4L sensor positions on the left

1 R, 2R, 3R, 4R sensor positions on the right

34 centre of mould in width direction

36 symmetry axis

38 meniscus level measured by camera

40 meniscus level measured by ultrasound sensor

42 standard deviation of meniscus level using single control

44 standard deviation of meniscus level using two symmetric control sensors

5 standard deviation of mould level using optical measurements using single sensor control

46 standard deviation of meniscus level by optical measurement using two symmetrical control sensors

48 standard deviation of stopper position during single sensor control

50 standard deviation of stopper position during two sensor control

52 power spectrum of stopper position

54 graph of mould level fluctuations over time

60-70 method steps.

Any reference signs in the claims should not be construed as limiting the scope of the appended claims.

Claims

1. A method of assessing the position of at least one mould level sensor (10) in the mould of a continuous slab caster (1) of molten metal, in particular molten steel, wherein the method is carried out in a caster arrangement (1) comprising a mould (2) having a width (MW) and a thickness (MT), the width being greater than the thickness, a submerged entry nozzle (3) through which a liquid is fed into the mould (2), wherein the submerged entry nozzle (3) is positioned at least approximately at the centre (34) of the mould in the width direction, and a valve (5) which regulates the flow of liquid through the submerged entry nozzle (3), the method comprising the steps of:

(a) arranging at least one mould level sensor (10) to measure the meniscus level (4) of the liquid at a first position along the width of the mould, thereby generating at least one meniscus level signal;

(b) operating the caster arrangement (1) while controlling the valve (5) in a feedbackloop in response to the at least one meniscus level signal;

(c) measuring the valve position and/or measuring the meniscus level (4) at one or several positions (1 L, 2L, 3L, 4L, 1 R, 2R, 3R, 4R) along the width of the mould, while operating the caster arrangement for a pre-determined time interval;

(d) calculating a measure for the fluctuation of the valve position and/or the meniscus level measured in step (c) over the pre-determined time interval;

(e) arranging at least one mould level sensor (10) to measure the meniscus level of the liquid at another position, different from the first position, along the width of the mould, thereby generating at least one further meniscus level signal, and repeating steps (b) to (d) using the at least one further meniscus level signal to control the valve (5); and

(f) determining an optimized position for the at least one mould level sensor (10) by selecting the position along the width of the mould, at which the measure calculated in step (d) is minimum.

2. The method of claim 1 , wherein the at least one mould level sensor (10) comprises at least a first mould level sensor (10a) and a second mould level sensor (10b), the at least first and second mould level sensors being positioned at different positions (1 L, 2L, 3L, 4L, 1 R, 2R, 3R, 4R) along the mould width, and wherein the valve (5) is controlled in a feedback-loop in response to a combination, in particular a mean, of the signals of the at least first and second mould level sensors (10a, 10b).

3. The method of claim 2, wherein the first and second mould level sensors (10a, 10b) are arranged on opposite sides of the submerged entry nozzle (3) along the width direction (MW) of the mould, and in particular are positioned at, at least approximately, equal distances from the centre of the mould (34).

4. The method of any one of the preceding claims 1 to 3, wherein the meniscus level (4) is measured in step (c) at several positions (4L,4R) spaced over the width (MW) of the mould, and wherein the measure for the fluctuation of the meniscus level is a combination, in particular a mean, of the fluctuation of the meniscus level (4) at each of the several positions.

5. The method of any one of the preceding claims 1 to 4, wherein the meniscus level (4) is measured in step (c) using a camera (32).

6. The method of any one of the preceding claims 1 to 5, wherein the measure for the fluctuation of the valve position and/or the meniscus level is calculated for a series of smaller time periods within the pre-determined time interval, and the measure for the fluctuation includes a number of peaks (56) in the measure for the fluctuation observed in the series of smaller time periods (54).

7. A system for controlling the meniscus level of molten metal, preferably molten steel, in the mould (2) of a continuous caster (1), the continuous caster comprising a mould (2) having a width (MW) and a thickness (MT), the width being greater than the thickness, a submerged entry nozzle (3) through which molten metal is fed into the mould, wherein the submerged entry nozzle (3) is disposed at least approximately at the centre of the mould in the width direction, and a valve (5) which regulates the flow of molten metal through the submerged entry nozzle (3), the system comprising at least a first mould level sensor (10a) and a second mould level sensor (10b) for measuring the level of molten metal in the mould (2), positioned at different positions along the width direction of the mould; a control unit (28) configured for receiving at least first and second meniscus level signals from the at least first and second mould level sensors (10a, 10b), respectively, calculating a combination, in particular a mean, of the at least first and second meniscus level signals and controlling the valve (5) in a feedback-loop in response to the combination of the meniscus level signals; and wherein the positions of the at least first and second mould level sensors (10a, 10b) have been determined by a method according to any one of claims 1 to 6.

8. The system of claim 7, wherein the first and second mould level sensors (10a, 10b) are arranged on opposite sides of the submerged entry nozzle (3) along the width direction of the mould, and in particular are positioned at, at least approximately, equal distances from the centre of the mould (34), in particular at distances from 9% to 20%, preferably from 12% to 17%, of the mould width.

9. The system of claim 7 or 8, wherein the at least first and second mould level sensors (10a, 10b) are disposed moveable along the width direction.

10. The system of claim 9, wherein the width of the mould (2) is configured to be adjustable during a continuous casting process, and the first and second sensors (10a, 10b) are configured to be automatically movable during a continuous casting process to positions along the mould width, in particular to positions on opposite sides of the submerged entry nozzle (3), which have been determine as optimized for the parameters of the continuous casting process, in particular for the current mould width.

11. The system of any one of claims 7 to 10, wherein the control unit (28) is configured to filter out a characteristic frequency of meniscus level fluctuations from the signal used for controlling the valve (5).

12. Use of a system according to any one of claims 7 to 11 in a continuous slab caster (1) of molten metal, in particular molten steel.

13. A computer program for implementing the method according to any one of claims 1 to 7, the computer program comprising computer-executable code which, when executed by a computer, performs the steps of: receiving a plurality of sets of values representing the valve position and/or the meniscus level at one or several positions along the width of the mould, measured over a pre-determined time interval; calculating measures for the amount of variation of each of the sets of values; determining the set of values having the lowest measure for the amount of variation as optimal.